Instructions to use GoedelMachines/diffusiongemma-26B-A4B-w4a16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use GoedelMachines/diffusiongemma-26B-A4B-w4a16 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="GoedelMachines/diffusiongemma-26B-A4B-w4a16", trust_remote_code=True)
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("GoedelMachines/diffusiongemma-26B-A4B-w4a16", trust_remote_code=True)
model = AutoModelForMultimodalLM.from_pretrained("GoedelMachines/diffusiongemma-26B-A4B-w4a16", trust_remote_code=True, device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use GoedelMachines/diffusiongemma-26B-A4B-w4a16 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "GoedelMachines/diffusiongemma-26B-A4B-w4a16"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "GoedelMachines/diffusiongemma-26B-A4B-w4a16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/GoedelMachines/diffusiongemma-26B-A4B-w4a16

SGLang

How to use GoedelMachines/diffusiongemma-26B-A4B-w4a16 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "GoedelMachines/diffusiongemma-26B-A4B-w4a16" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "GoedelMachines/diffusiongemma-26B-A4B-w4a16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "GoedelMachines/diffusiongemma-26B-A4B-w4a16" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "GoedelMachines/diffusiongemma-26B-A4B-w4a16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use GoedelMachines/diffusiongemma-26B-A4B-w4a16 with Docker Model Runner:
```
docker model run hf.co/GoedelMachines/diffusiongemma-26B-A4B-w4a16
```

diffusiongemma-26B-A4B-w4a16 / run.py

hsam007

Add files using upload-large-folder tool

5365e75 verified 2 days ago

Raw

History Blame Contribute Delete

1.07 kB

	"""run.py — load the W4 checkpoint with the fused kernels and generate.
	Usage: python run.py [your prompt...] (GB10/CUDA box; first forward JIT-compiles ~45s)
	"""
	import pathlib
	import sys
	import time

	HERE = pathlib.Path(__file__).resolve().parent
	sys.path.insert(0, str(HERE / "kernels"))

	from load_w4_checkpoint import load_fast

	CKPT = HERE if (HERE / "w4_meta.json").exists() else HERE / "ckpt"
	model, tok = load_fast(str(CKPT)) # tier="bitexact" for token-identical output

	prompt = " ".join(sys.argv[1:]) or "What is 17*23? Show your work."
	ids = tok.apply_chat_template([{"role": "user", "content": prompt}], add_generation_prompt=True,
	return_tensors="pt", return_dict=True)["input_ids"].cuda()
	t = time.time()
	out = model.generate(input_ids=ids, max_new_tokens=512)
	seq = out.sequences[0] if hasattr(out, "sequences") else out[0]
	gen = seq[ids.shape[1]:]
	print(tok.decode(gen, skip_special_tokens=True))
	print(f"\n[{len(gen) / (time.time() - t):.0f} tok/s — includes ~45s JIT on the first run]")